The SOL project is Twig Farm's LETR team NIA and Multi-campusIt is a program that works hand in hand to support the development of excellent artificial intelligence talents. This is an opportunity for trainees to experience practical business projects and experience the LETR team's research and development culture up close.
In the first phase of the project, 9 prospective developers have been with the LETR team for over 3 months since August. In order to experience the actual scene more closely, we were divided into 3 groups to work on a team project. I also had time to interact and collaborate with LETR team researchers in the industry through project mentoring.
This time, I listened to the stories of 9 participants in the first round, which finished a while ago. I think they all felt differently while actually working on the project. He shared an honest story about what he learned, thought, and felt through this experience.
In the first order, I would like to introduce the stories of 3 people who participated in the 'Natural Korean' project: Yechan, Chan Seon, and Jiwon.
Artificial intelligence is the first (?) Because
Hallo Please introduce yourself!
Praise: Hallo Originally, I majored in aviation maintenance engineering at Korea Aviation College, but I became interested in artificial intelligence while preparing for a new job.
Chan Seon: I am currently studying in the Department of Language and Cognitive Sciences at Hankuk University of Foreign Studies and am currently working as an intern at Twig Farm.
support: I graduated from Korea University's Department of Philosophy at the age of 28 this year. Until now, I was a non-major who lived a life dedicated to philosophy. Currently, I am very interested in artificial intelligence and logic.
Why did you become interested in artificial intelligence, and natural language processing in particular?
Praise: Above all, I thought it was a huge field that could be incorporated into various technologies. Historically, writing has accelerated development in many fields, and I also felt that NLP was an opportunity to challenge various fields.
Chan Seon: It started when I built up my linguistic skills at school. Personally, I was very interested in other people's perceptions of languages depending on the culture, which led me to start studying.
support: I've been studying philosophy for a long time, and I'm also interested in linguistics, so I thought, “Why is natural language intuitive?” I heard the question. In fact, languages must be understood intuitively, but as adults, it became very difficult to intuitively understand foreign languages.
In the case of artificial intelligence, I think deep learning-based translation is similar to intuitive understanding, and I think it is possible to understand human intuition in reverse through this. This part was interesting, and I became interested because I wanted to know the process.
After experiencing the SOL project
What led you to apply for the SOL project?
Praise: I wanted to experience how work is carried out in the business, what kind of data is used, and how data is built.
Chan Seon: In practice, I want to broaden my horizons by watching how natural language processing is being carried out and what fields and topics it is developing into.
support: I wanted to quickly understand how work is carried out in practice. And I wanted to grow faster by working on actual projects.
What was your experience with the SOL project?
Praise: We collected and analyzed data from previous papers, searched for necessary technologies, constructed the necessary data, compared and tested the model of the paper, and selected the model to be applied to the project. In particular, I remember having frequent discussions to define the project topic and set the direction. In the process, I was able to determine the direction of how to use what data, and set a framework for how to modify the model.
Chan Seon: In fact, the deep learning model was very difficult since it was almost my first time, but thanks to the team members who worked with me, I was able to improve my understanding. In the process, I think I was able to grow more than before by collecting data, pre-processing, etc.
support: The goal of the project was to create a kind of translator that converts unnatural Korean into natural Korean. Although there were realistic limitations, we trained reverse translated Korean and original Korean through various deep learning models.
What did you feel while working on a project with your team members?
Praise: Until now, I've thought it was natural to try and make a decision, and then set the direction according to the results. However, I was able to realize that this approach can be inefficient and that we have to change flexibly according to the situation.
Above all else, I felt that communication with team members was the most important thing. It's a pity that it would have been better if the project topic was defined more carefully and the direction of progress had been fully discussed. If I do a team project again, I would like to try to share opinions and interact more often.
Chan Seon: I thought it wasn't enough, so I tried to help the team even a little bit. Meanwhile, I felt that this field is developing a lot and has great potential in the future.
support: Personally, I learned that when running a baseline code, I should not only check what “works well,” but also identify the differences between the baseline code and my data. Otherwise, it's hard to figure out where the errors or issues that occur after running the baseline code come from.
What was the most memorable thing you did when finishing a project?
Praise: I was very excited because it was the first project we worked on together, and it was very helpful to have the experience of completing a project based on a paper. In particular, when I started working as a practitioner later, I was able to feel confident that I would be able to get through well based on this experience.
Chan Seon: What I remember most is that the team members helped each other and worked tirelessly together (although it was difficult).
support: I remember most about announcing the results of the project. As I organized the presentation materials, many thoughts went through my head. It wasn't a perfect result, and there were a lot of deficiencies, so I actually didn't feel good about it. However, the encouragement of the team members, the mentoring CEO, and the researchers helped me a lot.
The SOL Project and beyond
What fields do you want to challenge in the future, and what goals do you want to achieve?
Praise: Right now, I'm exploring various fields, but I have a dream of someday developing artificial intelligence like Iron Man's Jarvis. Specifically, I would like to develop a sub-flight system that can be mounted on an aircraft to communicate with the pilot and share and control information.
Chan Seon: Personal greed would like to contribute to the process of solving problems related to the linguistic significance of the Korean language. If possible, I would like to do research that can solve even mistranslation due to differences in nuance.
support: First, the goal is to finish project tasks that haven't been completed yet. Right now, I'm focusing on improving my coding skills, but I definitely want to achieve real results with this challenge.
Finally, please leave a message for the juniors who will experience the project in the future.
Praise: I think it would be nice to clearly set standards and proceed with the project through lots of conversations and discussions among team members. Additionally, I think it would be better if the COVID-19 situation improves and offline collaboration becomes more active.
Chan Seon: I did it too! You can do it too!
support: Time management is important. A week seemed to go slowly, but 3 months passed quickly. It's important to plan well and move regularly!
Good content to watch together
Why does LETR, a language processing engine, focus on text languages?
